METR publishes an assessment using embedded access
Why I called it
Equipment, office access and publication rights remove practical barriers to outside evaluation. A public report would demonstrate that the arrangement produces evidence readers can inspect; an access announcement alone cannot.
The call, in full. METR publishes a substantive Anthropic model assessment explicitly using its embedded access by January 31st 2027.
Scoring criterion. RESOLVES CORRECT if by 2027-01-31 23:59 UTC METR publishes on metr.org a report with substantive findings about an Anthropic model that explicitly attributes evidence or methods to the access announced October 15th 2026. Announcements without findings and assessments using only public API access do not count. RESOLVES WRONG otherwise.
The criterion is the machine-checkable version: a prediction that cannot be settled by a third party against a public source fails the build before it reaches this page.
Related
- Two more lab spin-outs of the Discovery Loop shape · 75%, open
- Palantir hits its raised FY2026 guidance · 80%, open
- A major breach names prompt injection as a root cause · 75%, open